Interframe Television Coding Using Gain and Displacement Compensation

Television signals contain a significant amount of frame-to-frame redundancy because of the 60-Hz field rate used to eliminate flicker. Some of this redundancy can be removed by techniques of conditional replenishment, 1 " 5 in which only picture elements that have changed by at least a threshold amount are transmitted. Conditional replenishment has recently been improved by displacement compensation.*5"1,3 The displacement compensation schemes have been developed primarily to compensate for translational displacement of objects in uniform illumination. This is done by estimating the translation of an object in the scene and using it for predictive coding by taking differences of elements with respect to appropriately displaced elements in the previous frame. Such schemes have been developed in 1227 both the picture element (pel) domain10,11 and the transform domain. 12,13 In this paper, we extend the translational displacement model of the picture intensity and develop recursive algorithms for estimating the parameters associated with the extended model. The extended model incorporates spatial and temporal variations of illumination, as well as translational displacement of moving objects. Illumination acts as a multiplicative factor or gain on the reflectance of objects in the scene. The extended model thus has two parameters, gain and displacement. The coder estimates these two parameters recursively from the previously transmitted data so that no additional information need be transmitted to specify them.